Research & Projects

Abstract

A substantial portion of household income in a developing country is spent on food consumption and thus a thorough understanding of global food price fluctuations in recent years convey crucial importance to bolster food and nutrition security in economically vulnerable nations. The main objective of this study is to determine the direction and magnitude of price and volatility transmission from international to domestic markets. We utilize Multivariate Generalized Autoregressive Conditional Heteroskedasticity framework on monthly price data spanning from January 2003 to November 2022 of major consumed agricultural commodities covering 42 developing countries in the world. Among the 75 sample markets, volatility spillover from the own markets is statistically significant for 71 tested markets; volatility spillover from international to domestic markets are statistically significant for 21 tested markets; and asymmetric effects are being statistically significant in 19 tested markets. The results of this research may provide a valuable insight for predicting agricultural commodity prices, enabling governments to support policy options that mitigate the impact of food grain price volatility and protect economic vulnerable groups from its adverse effects.

Abstract

The Dynamic Land Ecosystem Model (DLEM) Version 4.0 is our most recently updated spatially explicit, process-based, global scale land surface model that couples ecosystem dynamics to the cycles of water, carbon, nitrogen and phosphorus. Here, we discussed the several permafrost related processes and included the physical effects of those key equations into current DLEM version. To have an assessment on the model performance, we used DLEM to conduct historical daily soil temperature profile of a single gride cell at the West Siberian region in 1997 to simulate the annual freeze-thaw circle, and then performed the statistical analysis on different layers. Our results indicated the revised equations in particular the new moss layer did mitigate model simulation bias regarding thermal states with permafrost region compared to current version, however, further research and the ability to better represent the soil profile of high-latitude regions is still required. 

Project Background

Firms used to simply choose large companies as business partners because the size of companies usually means long-time good decisions, stable currency chain and high-level credit. However, as big booms happen through the internet, business related information on companies can be reached not only through financial reports or official statements on strategy but also from online communities or social media. Nowadays firms are more willing to weigh their vendors through different categories: financial situation, operational level strategy and supply chain evaluation. This project aims to help clients identify the risk of targeted vendors.

Problem Definition

To Identify the third-party risk, this project develops risk level classification models with financial data and NLP sentiment analysis with text data. The risk level classification model could predict risk level for targeted vendors based on the financial statements, mostly numerical data. In addition, the feature importance for risk level classification models is a great basis for clients to judge the third-party risk. NLP sentiment analysis, on the other hand, could reflect the reputation of the targeted vendors from public opinions and financial news.